JP2018536487A - Method for calibrating an X-ray image - Google Patents

Abstract

The invention relates to an X-ray device (1) in which X-rays (3) produced by an X-ray source (4) are emitted through an object (2) and recorded using an X-ray detector (5). It relates to a method for calibrating at least one two-dimensional X-ray image (31) of an object (2) to be imaged recorded by. X-ray source relative to the object (2) and / or relative to each other by comparing the already existing three-dimensional model (7) of the structure (6) of the object (2) with the two-dimensional X-ray image (31) The actual image positional relationship (15, 17) of (4) and the X-ray detector (5) is determined with respect to the two-dimensional X-ray image (31). [Selection] Figure 1

Description

  The present invention relates to a method for calibrating an X-ray apparatus for measuring a two-dimensional X-ray image of an object to be imaged, wherein X-rays produced by an X-ray source are emitted through the object and using an X-ray detector. Each two-dimensional X-ray image is recorded in that it is recorded, and the display of the object in the two-dimensional X-ray image is defined by the actual image positional relationship of the X-ray source and X-ray detector relative to the object. To do.

  Several methods for calibrating X-ray devices for three-dimensional measurement, such as X-ray CT devices or DVT X-ray devices, are known from the state of the art.

  DE 102008035412 discloses a method for compiling a three-dimensional X-ray image of the teeth of at least one sub-area of a subject, with a plurality of projections during a trajectory around the subject. Compile as a 3D X-ray image from the image. Prior to compiling the X-ray image, at least a portion of the object is shown as a visual display to know the relative position of the visual display relative to the current position of the device and the patient. Volume that is imaged in the visual display by overlaying the imaged volume at least approximately the correct position in the visual display, which is a function of the positioning of the object relative to the device and the selection of setting data and / or control data When the position and / or size of the image is changed, setting data and / or control data for generating a three-dimensional X-ray image is determined. The imaged volume is overlaid in the visual display only roughly as an approximate area. The recording angles of the two images can be different. The visual display can also be an existing 3D image, such as a 3D X-ray image. Position registration is performed so that the volume to be imaged is correctly positioned in the visual display, and in order to correctly display the position of the imaged volume in the visual display, the relative position of the device to the patient is: Compare with the device in the visual display and the current position of the patient.

  US Patent Application Publication No. 2011/0255765 discloses a system and method for removing artifacts from an x-ray image of a patient's teeth, the system comprising a surface scanner that generates a surface image of the patient's teeth. The surface data of the surface image and the volume data of the CT scan are aligned and properly positioned and superimposed on each other to produce a combined data set. The volume data data points that extend beyond the tooth surface into the surface data are identified and removed, resulting in the removal of artifacts created by the metal part.

  In one embodiment, the surface model of the surface data is projected in the same two-dimensional format as the CT projection data to produce forward projection data. Forward projection data is combined with CT projection data to identify artifacts. There is no need to manually pre-position the two sets of projection data relative to each other.

  In the DVT method, an X-ray tube and a surface digital image sensor rotate around an object and generate a plurality of two-dimensional X-ray images from different recording angles on a partial trajectory of the X-ray tube. Subsequently, a reconstruction method is used to reconstruct a 3D X-ray image from individual 2D X-ray images. The correct imaging geometry or image positional relationship of the X-ray tube and X-ray detector relative to the object is important for the reconstruction method. Usually, a calibration phantom is used to define this image positional relationship. Such calibration measurements are performed when the X-ray device is still in the factory and periodically after the first operation of the X-ray device.

  The disadvantage of this method is that the original positional relationship defined during calibration can change, for example, as a result of wear by the mechanism of the X-ray device or by changing the frictional force in driving the X-ray device. is there. As a result of the change in the positional relationship, the actual positional relationship between the X-ray emitter and the X-ray detector relative to the object does not match the predetermined positional relationship derived from the calibration measurement. As a result, artifacts such as distortion, detail, streaks and / or shadow blur can occur in the reconstructed three-dimensional X-ray image. Motion artifacts can also occur when a patient moves during imaging.

  Inaccurate image positional relationships also cause additional errors in the X-ray image of the device that generates the panoramic tomographic image, or in the oral X-ray image that depends on the known calibrated positional relationship.

  Therefore, it is an object of the present invention to provide a method for calibrating an X-ray device that ensures that an object is measured without error without using a calibration phantom.

  The invention relates to at least one two-dimensional image of an object to be imaged, recorded by an X-ray device, in which X-rays produced by an X-ray source are emitted through the object and recorded using an X-ray detector. The present invention relates to a method for calibrating an X-ray image. An already existing 3D model with the structure of the object is compared with a 2D X-ray image and the actual image positional relationship of the X-ray source and X-ray detector relative to the object and / or relative to each other is The determination is made for a dimensional X-ray image.

  A two-dimensional X-ray image can be recorded using, for example, a DVT X-ray device. Therefore, a plurality of two-dimensional X-ray images are generated from different recording angles in the course of one orbit. The x-ray source generally rotates 180 ° to 360 ° around the object, producing an x-ray cone fan that is typically pulsed. The X-rays pass through the three-dimensional object and produce a weakened grayscale X-ray image as a two-dimensional X-ray image on the X-ray detector for each respective recording angle or each recording time point.

  A comparison between an existing 3D model and a 2D X-ray image can be performed, for example, by assigning structural points in the 3D model to corresponding structural points in the 2D X-ray image.

  Store the determined actual image positional relationship of the X-ray source and X-ray detector relative to the object or relative to each other for later use in X-ray image correction, reconstruction, optimization and / or calculation. obtain.

  Thus, the calibration method needs to be performed once or at defined intervals to verify the calibration.

  One advantage of this method is that it is not necessary to first calibrate with a calibration phantom to measure the object. This reduces maintenance requirements.

  Another advantage of the method is that calibration is performed on each individual two-dimensional X-ray image by comparison with a three-dimensional model from a specific recording angle, so that the change in image positional relationship is reduced. The resulting unexpected malfunction of the X-ray device is corrected immediately.

  Another advantage of the method is that it can generate a three-dimensional X-ray image of a patient who is moving while in orbit. For example, the entire mandible can be clearly reconstructed as an imaging structure without movement artifacts, even when the head moves sideways or the patient is chewing. This is the exact image position of this structure, eg, the lower jaw, relative to the X-ray source and X-ray detector, and the individual 2D X-ray image and the 3D model in the virtual 2D projection image. This is because the determination is made by comparing the virtual projection. Calibration for the structure of the object to be displayed, such as the lower jaw, is therefore determined in the best possible manner regardless of patient movement.

  When comparing a three-dimensional model with a two-dimensional X-ray image, a virtual projection method is applied to at least a part of the three-dimensional model to generate a virtual two-dimensional projection image by considering a predetermined image positional relationship; The structure in the two-dimensional X-ray image is compared with the structure in the virtual two-dimensional projection image.

  Therefore, in the virtual projection method, the imaging geometry of each two-dimensional X-ray image is simulated using a computer, thereby generating a corresponding virtual two-dimensional projection image. In the projection method, the image positional relationship between the virtual X-ray source and the virtual X-ray detector with respect to the three-dimensional model of the structure is simulated as such, and the X-ray in the form of a cone fan passes through the three-dimensional model. The method of displaying the three-dimensional model on the virtual X-ray detector is simulated. As a result, the shape of the structure image in the two-dimensional X-ray image matches the virtual projection of the structure in the virtual two-dimensional projection image with the same image positional relationship.

  In an alternative method, a virtual backprojection method can be applied to a two-dimensional X-ray image to perform a comparison between the three-dimensional model and the two-dimensional X-ray image. Displayed in space. This projection is then compared to an existing 3D model.

  Comparing the structure in the two-dimensional X-ray image with the virtual two-dimensional projection image, the deviation is identified, and within the framework of the optimization method, the image positional relationship is gradually changed so that the deviation is smaller than the defined threshold value. Until then, a new virtual two-dimensional projection image is generated after each change.

  The predetermined image positional relationship of the x-ray source and x-ray detector relative to the object can be based on calibration or subsequent calibration, or can be determined mathematically based on a model. Therefore, the predetermined image positional relationship is an initial solution, and based on this, the actual image positional relationship is determined using an optimization method. A so-called Kalman filter can be used in the optimization method, which is used to stabilize the initial solution. A Kalman filter is used to eliminate interference caused by the measuring device.

  When the threshold value is not reached, the structure in the two-dimensional X-ray image is defined so as to correspond to the virtual two-dimensional projection image as much as possible, and as a result, the image positional relationship determined after the optimization method is finished. Corresponds to or at least similar to the actual image positional relationship of the respective two-dimensional X-ray image with respect to the structure of the object to be displayed.

  The comparison can be performed using a comparison operator. At that time, the comparison operator is optimized by an optimization method. The optimization method may be completed after reaching a predetermined threshold, for example. The threshold that is the quality accuracy of the optimization may be, for example, 10% that is the calibration accuracy of each X-ray apparatus.

  The optimization method can either optimize the comparison operator directly or optimize the deviation between the virtual 2D projection image and the structure in the 2D X-ray image.

  While the X-ray source and X-ray detector are moving around the object, two-dimensional X-ray images are advantageously recorded one by one from different recording angles, using a reconstruction method, By grasping the determined actual image positional relationship of the image, the target three-dimensional X-ray image or the target panoramic tomographic image is generated from the recorded two-dimensional X-ray image.

  The motion of the x-ray source and x-ray detector about the object can be a circular rotation in the form of a partial trajectory surrounding the object, or the orientation of the x-ray source and x-ray detector relative to the object. And by changing the position, it can be a different movement.

  As a result, the method can be used to calibrate a DVT X-ray device or a CT X-ray device. For each individual two-dimensional X-ray image, the calibration of the image positional relationship is verified and corrected as such. Thus, this allows dynamic verification, so that even a sudden mechanical failure of the device is taken into account. In the reconstruction method, the grayscale image in the two-dimensional X-ray image corresponds to the total absorption along the measured X-ray path through the object. At that time, the X-ray path is broken down into small voxels. In backprojection, the measured values along each measured X-ray path are distributed as well as possible to the voxels placed along the path. For 2D X-ray images, this is done from various recording angles, so that an excellent prediction of the 3D X-ray image to be displayed is obtained as a result.

  For reconstruction of the panoramic tomographic image, a computer is used to calculate the panoramic tomographic image from two-dimensional X-ray images taken from different recording directions.

  Therefore, the initial solution is a planned path of the X-ray source and X-ray detector centered on the object, and may have been taken from, for example, factory calculations. For each recording angle, this path connecting the individual positions of the X-ray source and X-ray detector relative to the object is then optimized or gradually optimized until an actual path is determined. More accurately defined in the course of the method. The actual path is then used for reconstruction, and as a result, an error-free 3D X-ray image or panoramic tomographic image is reconstructed.

  The structure of the object may advantageously be the upper jaw, the lower jaw, a group of teeth, a dental prosthesis, a filling, an inlay, the entire object, a part of the object, the patient's head, and / or individual teeth.

  Thus, the subject can also be a part of the jaw consisting of a plurality of teeth and a dental prosthesis.

  Existing three-dimensional models of structures can be advantageously recorded using optical three-dimensional surface measurement methods, which include only one surface of the structure. During the comparison, the surface edge of the structure in the two-dimensional X-ray image is then compared with the surface edge of the structure in the virtual two-dimensional projection image.

  An existing three-dimensional model of the structure may comprise the entire surface of the structure, or even only a portion of the surface of the structure. When measuring using a three-dimensional surface scanner, for example, only the visible surface of a tooth can be measured. The surface measurement method can be, for example, a fringe projection method, a confocal measurement method, or a laser scanning method. The surface edge of the structure, which can be compared with the surface edge of the structure in the two-dimensional X-ray image, therefore appears in the virtual projection of the three-dimensional model. The tooth edges, i.e. even the tooth depressions clearly visible in the two-dimensional X-ray image, for example, can be used for comparison.

  Existing 3D models may also only depict points that represent the surface or characteristic points of the structure.

  An existing 3D model of the structure is advantageously generated by recording the impression of the structure, utilizing optical 3D surface measurements, the 3D model includes only one surface of the structure, The surface edge of the structure in the two-dimensional X-ray image is compared with the surface edge of the structure in the virtual two-dimensional projection image.

  As a result, only one impression of the structure is measured, so that the surface of the structure is determined therefrom. For example, the impression of one or more teeth can be measured.

  Existing 3D models of structures can be advantageously recorded using 3D volumetric measurements, in particular using MRI, CT or DVT methods, including volume data of structures, The structure in the dimensional X-ray image is compared with a projection that simulates the structure in the virtual two-dimensional projection image.

  As such, the three-dimensional model may also comprise volume data with information regarding the internal composition of the structure. Substructures within the structure, such as the separation surface between the tooth and the surrounding gingiva, or the shape of the root or jawbone may also be used for comparison.

  Magnetic Resonance Imaging (MRI) is based on the principle of nuclear magnetic resonance physically, and the separation surface between soft tissue and hard tissue, such as between teeth and surrounding gingiva, is clearer than three-dimensional X-ray images. Is displayed.

  Computed tomography (CT) uses a multi-row detector as an X-ray detector based on the reconstruction of a three-dimensional X-ray image from individual two-dimensional X-ray images of an object at different recording angles.

  In digital volume tomography (DVT), a three-dimensional X-ray image is similarly reconstructed from individual two-dimensional X-ray images with different recording angles, and a planar detector is used as the X-ray detector.

  Advantageously, in the projection method, not only the image positional relationship of the X-ray source and the X-ray detector relative to the object, but also the thickness of the structure to be imaged, and thus the X-ray attenuation and / or structure of the structure. Material is also considered for virtual projection. Therefore, X-ray attenuation depending on the structure is also considered. That is, the comparison of the virtual projection image and the two-dimensional X-ray image is not limited to the edge of the structure but also includes additional image content.

  Therefore, it is possible to improve the projection simulation in the virtual two-dimensional projection image by considering the thickness of the structure and the material. While measuring a structure using an optical surface scanner, it can be determined, for example, whether the structure is a natural tooth or a dental prosthesis made of ceramic, gold or plastic. The respective factors of X-ray attenuation for gold, plastic or ceramic are thereby taken into account.

  When performing the optimization method, a predetermined image positional relationship with known factory calibration may be advantageously used as an initial solution.

  As a result, the first solution that is already very close to the actual image positional relationship is used.

  As a result, the time and computational attempts required for the optimization method are reduced.

  When comparing a two-dimensional X-ray image with a virtual two-dimensional projection image, the similarity is advantageously calculated for the determination of the deviation, the gradient difference method, the direct difference method, the correlation method, the first rank and / or Use higher rank cross-correlation, statistical, or least square error methods.

  When using the mentioned method, for example, matching patterns in the two-dimensional X-ray image and the virtual two-dimensional projection image, such as the end of a tooth, are compared with each other to determine the actual image positional relationship. As the similarity of these patterns increases, the similarity increases. When the image positional relationships in the two-dimensional X-ray image and in the simulated two-dimensional projection image match, the patterns must match each other or at least partially resemble each other. Therefore, in this case, the degree of similarity is maximized, and as a result, the optimization method can be completed.

  The statistical method can be, for example, a so-called mutual information method.

  By using the optimization method, the similarity can be advantageously increased and the deviation can be decreased until the optimum and the relationship of the real image due to the optimization is determined.

  As a result, when using the optimization method, the image-related solution approaches the optimal solution.

  Changes in the image positional relationship during the optimization method can be advantageously described using a transformation matrix.

  By using the transformation matrix, the 3D model can therefore be virtually shifted or rotated relative to the X-ray source and X-ray detector virtually.

  With this method, this first structure can be advantageously selected in a first step of determining the first actual image positional relationship of the two-dimensional X-ray image with respect to the first structure of interest, In a second step of determining a second actual image positional relationship of the two-dimensional X-ray image with respect to the target second structure, the second structure is selected.

  Therefore, the corresponding actual image positional relationship for each selected structure is determined.

  The first three-dimensional X-ray image can be advantageously reconstructed using the first actual image positional relationship, and the second three-dimensional X-ray image can be reconstructed using the second actual image positional relationship. The first region in the first three-dimensional X-ray image that clearly displays the first structure, and the second three-dimensional X-ray that clearly displays the second structure. It is integrated with the second region in the image to form one entire three-dimensional X-ray image of the object.

  Therefore, as a result, at least the selected structure is clearly displayed in the corresponding three-dimensional X-ray image of this structure.

  Advantageously, the first structure can be the lower jaw or a part of the lower jaw and the second structure can be the upper jaw or a part of the upper jaw.

  Therefore, a first three-dimensional X-ray image is generated with the lower jaw as the selected structure, so that even if the lower jaw moves during recording, the lower jaw is clearly displayed while the upper jaw and head of the patient are displayed. The rest is blurred. Therefore, in the second 3D X-ray image, the upper jaw and the rest of the head of the patient are clearly displayed, while the lower jaw is blurred. The sharp regions of the two three-dimensional X-ray images can then be merged into one overall three-dimensional X-ray image that clearly shows both the lower and upper jaws.

The present invention will be described with reference to the drawings. The drawings show:
FIG. 1 is a sketch for explaining the present method. FIG. 2 is a sketch for explaining the optimization method for the first solution. FIG. 3 is a sketch for explaining the optimization method for the last solution.

  FIG. 1 shows a sketch illustrating the method for calibrating the X-ray device 1 for the measurement of a two-dimensional X-ray image of an object 2 to be imaged, such as a patient's head. Each two-dimensional, in that the X-ray 3 in the form of a conical fan produced by the X-ray source 4 is emitted through the object 2 and recorded using an X-ray detector 5 such as a flat panel detector. Record an X-ray image. Object 2 includes a structure 6 to be imaged, such as the lower jaw, shown in broken lines. A three-dimensional model 7 already exists for the imaged structure 6, i.e. the lower jaw, recorded using optical three-dimensional surface measurement. In this case, the existing three-dimensional model 7 includes only four molars 8 on the right side of the lower jaw 6. For example, a three-dimensional model can be recorded using a small three-dimensional dental camera based on a fringe projection method or a confocal measurement method. A three-dimensional model can also be generated by first taking an impression of the tooth 8 and then measuring this impression utilizing a miniature dental camera. Therefore, the three-dimensional model 7 includes only the visible surface of the molar 8. During the measurement of the three-dimensional X-ray image 9, the X-ray source 4 and the X-ray detector 5 rotate gradually around the rotation point 10 in the measurement volume, and the individual two-dimensional X of the object 2 and thus the structure 6. Line images are recorded from different recording angles 11 indicated by arrows, and a target three-dimensional X-ray image 9 is generated from two-dimensional X-ray images recorded from different recording angles 11 by using a reconstruction method. The sub-area 12 of the structure 6 to be imaged, i.e. the lower jaw, is indicated by a broken line in the three-dimensional X-ray image 9. A predetermined image positional relationship 13 of the X-ray source 4 is indicated by a solid line. Directly opposite thereto, the predetermined image positional relationship 14 of the X-ray detector 5 is similarly indicated by a solid line. The actual image positional relationship 15 of the X-ray source 4 with respect to each recording angle 16 is indicated by a broken line by shifting to that position. Similarly, the actual image positional relationship 17 of the shifted X-ray detector 5 is also indicated by a broken line. Therefore, the X-ray 18 in the form of a cone of cone beam is similarly shifted and records the object 2 from the actual recording angle 19 that deviates significantly from the predetermined recording angle 16. This deviation can be caused, for example, by a mechanism or drive resulting from alignment and by increased friction forces in the drive of the X-ray device 1. The method therefore serves to compensate for this deviation, indicated by arrow 20. Further, this deviation between the predetermined image positional relationship 13, 14 and the actual image positional relationship 15, 17 can also occur in the radial direction with respect to the rotation point 10. The individual image positional relationships 13 of the X-ray source 4 and the predetermined image positional relationships 14 of the X-ray detector 5 for all recording angles 11 then form an orbital path 21 indicated by a solid line. After performing the optimization method for each individual recording angle 11, the actual image positional relationship 15 of the X-ray source 4 and the actual X-ray detector 5 forming the actual trajectory path 22 indicated by the broken line. The image positional relationship 17 is determined. Deviations between the predetermined trajectory path 21 and the actual trajectory path 22 can be caused not only by the failure of the drive mechanism but also by patient movement during trajectory movement. Therefore, the original reconstruction using a predetermined trajectory path 21, which can be predefined by factory calibration, for example blur, which specifically displays a blurred image of the imaged structure 6 to include interference artifacts. 3D X-ray images are reconstructed. The reconstruction using the determined actual trajectory path 22 produces a three-dimensional X-ray image 9 that clearly and clearly shows details and the structure 6 to be imaged, for example the lower jaw. The three-dimensional X-ray image 9 and the three-dimensional model 7 are displayed using a display device 23 such as a monitor. Image data of the X-ray detector 5 such as individual two-dimensional X-ray images is transmitted to the computer 24 by cable or wirelessly. The reconstruction method and the optimization method are also performed using the computer 24. The computer 24 is connected to input means such as a keyboard 25 and a mouse 26 so that the user can operate using the cursor 27.

  FIG. 2 shows a sketch illustrating the optimization method. By applying the projection method to the three-dimensional model 7 from FIG. 1 and considering the predetermined image positional relationship 13 of the X-ray source 4 and the predetermined image positional relationship 14 of the X-ray detector 5 with respect to the recording angle 16, A virtual two-dimensional projection image 30 is generated. Therefore, after irradiating the three-dimensional model 7 with a virtual X-ray source and simulating X-ray absorption, the two-dimensional projection image 30 is displayed on the virtual X-ray detector, whereby the X of the virtual three-dimensional model 7 is displayed. Virtually simulate the irradiation. An actual two-dimensional X-ray image 31 from the actual recording angle 19 is displayed for comparison. In this case, only the sub-area 12 with the molars 8 is displayed. It can clearly be seen that the two-dimensional projection image 30 of the three-dimensional model 7 and thus the molar 8 deviates significantly from the associated two-dimensional X-ray image 31 in its form. This is caused by a deviation 20 between the predetermined recording angle 16 and the actual recording angle 19. In particular, the tooth end 32 can be seen in both the two-dimensional projection image 30 and the two-dimensional X-ray image 31 and can therefore be used in a comparison method. Characteristic structures such as dental cusps 33 and dental depressions 34 can also be used in the comparison method. For the comparison method, similarity is determined, which is a highly reliable measure of similarity between two structures. The comparison method can be automatically performed by utilizing the computer 24 of FIG.

FIG. 3 shows a sketch for explaining the optimization method. Compared with FIG. 2, the actual positional relationship 15 of the virtual X-ray source and the actual image positional relationship 17 of the virtual X-ray detector from FIG. Thus, the actual two-dimensional projection image 40 is generated by using the projection method. Compared to the two-dimensional X-ray image 31, it is now clear that the shape of the tooth end 32, tooth cusp 33 and tooth recess 34 of the tooth 8 match. The virtual projection image is calculated based on any deviation of the geometric image positional relationship, and therefore any deviation is between the structures in the virtual production image and the two-dimensional X-ray image. Leads to results with lower similarity. Therefore, the similarity reaches its maximum, so that the optimization method can be completed and the actual image positional relationship of the X-ray source and X-ray detector is determined. The optimization method proceeds step by step from the first solution in FIG. 2 to the final solution in FIG. The change 20 between the predetermined image positional relationship 13 and the actual image positional relationship 15 can be described using, for example, a transformation matrix.

Explanation of symbols
1 X-ray equipment
2 Object to be imaged
3 X-ray
4 X-ray source
5 X-ray detector
6 Structure to be imaged
6 lower jaw
7 Three-dimensional model
8 molars
9 Three-dimensional X-ray image
10 Rotation point
11 Recording angle
12 sub-areas
13 Predetermined image position relationship
14 Predetermined image position relationship
15 Actual image position
16 Recording angle
17 Actual image location
18 X-ray
19 Actual recording angle
20 arrows
20 deviations
21 Predetermined orbital path
22 Actual trajectory path
23 Display device
24 computers
25 keyboard
26 mouse
27 Cursor
30 Virtual 2D projection images
31 Actual two-dimensional X-ray image
32 Teeth end
33 Teeth
34 Tooth depression
40 Determined virtual 2D projection image

Claims (12)

  Recorded with X-ray device (1), where X-rays (3) produced by X-ray source (4) are emitted through object (2) and recorded with X-ray detector (5) A method for calibrating at least one two-dimensional X-ray image (31) of the object (2) to be imaged, comprising an already existing three-dimensional model (6) comprising the structure (6) of the object (2) 7) is compared to the two-dimensional X-ray image (31) and the actual X-ray source (4) and X-ray detector (5) relative to the object (2) and / or relative to each other. Image positional relationship (15, 17) is determined with respect to the two-dimensional X-ray image (31), the comparison between the three-dimensional model (7) and the two-dimensional X-ray image (31), and a predetermined In order to consider the image positional relationship (13, 14) of the three-dimensional model (7), a virtual projection method is used. Both are applied to a part to generate a virtual two-dimensional projection image (30, 40), and the structure (6) in the two-dimensional X-ray image (31) is converted into the virtual two-dimensional projection image (30, 40). During the comparison of the structure (6) in the two-dimensional X-ray image (31) with the virtual two-dimensional projection image (30, 40). Within the framework of the optimization method, a deviation is identified, and the image positional relationship (13, 14) is gradually changed until a new virtual two-dimensional projection image (after each change) until the deviation becomes smaller than a defined threshold. 40).   The two-dimensional X-ray image (31) is obtained from different recording angles (11) while the X-ray source (4) and the X-ray detector (5) are moving around the object (2). 3D X-ray image of the object (2) by recording each one and using the reconstruction method and grasping the determined image positional relationship (15, 17) of the 2D X-ray image (31) The method according to claim 1, characterized in that a panoramic tomographic image of the object (2) or the object (2) is generated from the recorded two-dimensional X-ray image (31).   The structure (6) of the subject (2) includes an upper jaw, a lower jaw (6), a group of teeth, a dental prosthesis, a filling, an inlay, the entire subject, a portion of the subject, a patient's head, and / or 3. A method according to claim 1 or 2, characterized in that it is an individual tooth.   The existing three-dimensional model (7) of the structure (6, 8) is recorded using an optical three-dimensional surface measurement method, and the three-dimensional model (7) is a surface of the structure (6, 8). During the comparison, the surface edge (33) of the structure (6, 8) in the two-dimensional X-ray image (31) is converted into the virtual two-dimensional projection image (30, 40). The method according to claim 1, characterized in that it is compared with a surface edge (33) of the structure (6, 8).   The existing three-dimensional model (7) of the structure (6) is generated by recording an impression of the structure (6, 8) using an optical three-dimensional surface measurement method, and the three-dimensional model ( 7) includes only one surface of the structure, and during the comparison, the surface edge of the structure in the two-dimensional X-ray image (31) is represented in the virtual two-dimensional projection image (30). The method according to claim 1, wherein the method is compared with a surface edge of the structure.   The existing three-dimensional model (7) of the structure (6) is recorded using a three-dimensional volume measurement method, in particular, using the MRI method, CT method or DVT method, and the volume data of the structure And comparing the structure in the two-dimensional X-ray image (31) with a simulated simulation of the structure in the virtual two-dimensional projection image (30) during the comparison. The method according to claim 1.   In the virtual projection method, not only the image positional relationship (13, 14, 15, 17) of the X-ray source (4) and the X-ray detector (5) relative to the object (2) is imaged. The thickness of the structure (6, 8) and thus the X-ray attenuation by the structure (6, 8) and / or the material of the structure (6, 8) and the X-ray depending on the structure The method according to claim 1, wherein source weakness is also taken into account.   8. The method according to claim 1, wherein the predetermined image positional relationship (13, 14) with a known calibration is used as the first solution for the execution of the optimization method. 9. the method of.   When comparing the two-dimensional X-ray image (31) with the virtual two-dimensional projection image (30, 40), the similarity is calculated for the determination of the deviation, and the slope difference method, direct difference method, correlation method, 9. A method according to any one of the preceding claims, characterized in that it uses a first rank and / or higher rank cross-correlation method, a statistical method or a least square error method.   The first structure (6, 8) of the object (2) is determined by this method for the first actual image positional relationship of the two-dimensional X-ray image (31) with respect to the first structure. The second structure selected in the first step is used to determine the second actual image positional relationship of the two-dimensional X-ray image (31) relative to the second structure by the method. 10. The method according to claim 1, wherein the selection is performed in two steps.   A first three-dimensional X-ray image (9) is reconstructed using the first actual image positional relationship, and a second three-dimensional X-ray image is used using the second actual image positional relationship. Reconstructing, and then displaying the first structure (6, 8) clearly, displaying the first region in the first three-dimensional X-ray image, clearly displaying the second structure, 11. A method according to claim 10, characterized in that it is integrated with a second region in a second 3D X-ray image into one overall 3D X-ray image of the object (2).   The method according to claim 10 or 11, wherein the first structure is a lower jaw or a part of the lower jaw, and the second structure is an upper jaw or a part of the upper jaw.

Images